Neoplasia involves the clonal expansion of cells which can escape the inhibitory influences that normally limit cell growth. The revolution in cancer research that has occurred over the last decade has largely involved the discovery of genes which confer these inhibitory influences and the delineation of mutations of these genes in various tumour types. TGF-.beta. is one of the cytokines that can inhibit epithelial cell growth, and has been extensively studied for over a decade (1-5). Resistance to TGF-.beta. is common in human cancers of many types, emphasizing the importance of this secreted polypeptide to the neoplastic process (6). However, the mechanisms by which tumour cells become resistant to TGF-.beta. are generally unknown.
Much has been learned about TGF-.beta., its receptors, and its physiologic effects. TGF-.beta. is now known to be the prototype for a large and conserved family of related polypeptides which have diverse physiologic functions in organisms as disparate as Drosophila, C. elegans, and Homo sapiens. The TGF-.beta. signal is initially received by a receptor complex containing the products of three different genes (TGF-.beta. receptor types RI, RII, and RII). This signal is apparently mediated by the serine/threonine kinase activities of RI and RII. Though a small number of potential substrates for these receptor kinases have been identified (7-11) the biochemical pathways that are stimulated by these kinase activities are largely unknown. However, genes which appear to function downstream of these receptors, on the basis of genetic criteria, have been identified in Drosophila and C. elegans. Mutations of the C. elegans genes sma-2, sma-3, and sma-4 confer phenotypic abnormalities identical to those observed with mutants of a TGF-.beta.-family receptor gene (daf-4) in this organism (12). Savage et al. have proposed the name "dwarfins" for the corresponding proteins, homologues of which have been identified in the human (12). Similarly, mutations in the Drosophila Mad gene result in phenotypes like those observed in strains with mutations of the TGF-.beta.-like ligand dpp (13,14). The Mad and sma genes are highly related by sequence, suggesting they all function similarly as mediators of TGF-.beta.-like signaling (13,14). The Mad and sma genes have no relationship to other known signaling molecules, and encode no motifs that provide clues to their biochemical function. It is likely, however, that the final arbiters of the growth inhibition conferred by TGF-.beta. family members include the cyclin-dependent kinase inhibitors (15).
The importance of the TGF-.beta. pathway to colorectal tumorigenesis has recently been highlighted by two observations. A subset of colorectal tumors has been shown to harbor inactivating mutations of the TGF-.beta. RII gene (16,17). This subset, accounting for about 15% of total colorectal cancers (18-20), comprises tumours with characteristic defects in mismatch repair (21), and the high frequency of TGF-.beta. RII gene mutations in this tumour type results from a mutation-prone polyadenosine tract within the coding region of the gene. Though most other colorectal cancers have been shown to be resistant to the inhibitory effects of TGF-.beta. (2,6), the cause of such insensitivity is unknown. Mutations of the TGF-.beta. RII gene or of the cdk inhibitor genes p15, p16, and p21 are not generally found in these tumours (22,23), suggesting that the defects lie in the intermediates in the signaling pathway. This hypothesis recently received experimental support with the discovery of the DPC4 gene (24). DPC4 was identified through a positional cloning approach designed to identify a pancreatic tumour suppressor gene on chromosome 18q21. The DPC4 gene was found to be highly related to Mad and sma and genetic alterations which affected DPC4 were observed in over 50% of pancreatic cancers. Analogous studies of other human cancers, including those of the colon, revealed that DPC4 was genetically altered in only a minority of cases (25,26). There is a need in the art for identification of additional human genes homologous to DPC4 and its lower eukaryotic homologues (hereinafter referred to as the Mad gene family) which are involved in neoplastic processes.